1. Field of the Invention
The present invention relates to a driving safety system and, more particularly, to a driving safety system and a barrier screening method.
2. Description of the Related Art
To keep abreast with technological progress, developing a smart driving system has gradually become a trend in the smart car market. Current smart driving systems identify a relationship between the driver's habitual safety distance and a car speed with information, such as car speed, distance behind a vehicle and the like, analyzed by a neural network for the driver's driving habits, such that the safe distance is configurable depending on the car speed. The stability of those smart driving systems is in turn closely bound up to failure of the smart driving systems. Besides, those driving systems are prone to influence of complicated environment or noise, which can lead to errors in detecting a front target vehicle, especially in a turning operation, and result in issues of system malfunction and lower system effectiveness.
Current driving systems, such as Volvo's City Safety system, operate when the vehicle is within a speed range 1530 km/h. When the system detects a front barrier, the City Safety system will slow down and then fully stops the vehicle if its speed is under 15 km/h. One of the current driving systems, Mazda's SCBS system, employs a laser detector to capture information of a barrier in the front and detects a distance to the barrier in order to determine whether the host car desires a braking control to decelerate the vehicle and lower the chance of collision or not when the car speed is within a range 4˜30 km/h. Ford has introduced a collision avoidance system as an essential safety system equipped in many models thereof, which begins active brake control when the car speed is within 5˜30 km/h and a detected distance to a front barrier is less than an alert value. It is applicable to a curved road with the radius of a curvature greater than 20 m.
As disclosed in Taiwan Patent No. I318604, entitled “Method for estimating time to collision using recursive least square algorithm”, the method targets at enhancing accuracy in calculating a time to collision, and adopts a distance detector mounted on a vehicle to measure a relative distance between the vehicle and an external vehicle or a barrier. An estimation unit mounted on the vehicle reads the relative distance measured by the distance detector, and it calculates a curve of a second order with the relative distance using the recursive least square (RLS) method to substitute multiple known parameters into equations associated with the curve and the RLS method for obtaining a time point when the relative distance is zero and for estimating time to collision and a time difference therebetween. Thus, despite noise interference, the negative effect of noise can be mitigated to avoid the chance of collision.
Given the foregoing smart driving systems and method, the vehicle speed can be decelerated, the chance of collision can be lowered, while the following drawbacks fail to be ruled out.
1. Vehicles have to be driven at a speed under 30 km/h, and the application is only applicable to certain driving circumstances with the conventional smart driving systems.
2. A braking timing or an advance alert fails to be accurately determined or provided to result in a slower vehicle speed in an attempt to prevent a late braking control.
3. Current techniques lack a prediction mechanism and enough accuracy and thus lead to unstable systems and accidents arising from frequent troubles in the systems.
Moreover, to increase the accuracy of the systems in determining a barrier, conventional techniques estimate the time to collision and the time difference with complicated equations to lessen the noise effect and avoid the occurrence of collision. However, lots of time and computational resources are required, and situations to be screened out are mostly involved in those having a floating relative distance between the vehicle with the driving system and an external vehicle or a barrier, because no determination is made whether a target in the front actually exists and whether the target in the front is directly assumed to be an actually existing entity.